Reexamining the relationship between Apollinaris Patera and the basalts of the Gusev crater plains, Mars

Transcription

1 Click Here for Full Article JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115,, doi: /2009je003397, 2010 Reexamining the relationship between Apollinaris Patera and the basalts of the Gusev crater plains, Mars Nicholas P. Lang, 1 Harry Y. McSween Jr., 2 Livio L. Tornabene, 3 Craig J. Hardgrove, 2 and Philip R. Christensen 4 Received 10 April 2009; revised 26 October 2009; accepted 10 November 2009; published 16 April [1] A possible source of the Gusev crater basalts analyzed by the Spirit rover is Apollinaris Patera. We test this hypothesis by identifying and analyzing potential lava flow paths using Thermal Emission Imaging System, Mars Orbiter Camera, High Resolution Imaging Science Experiment, and Mars Orbiter Laser Altimeter data sets together with published geologic mapping. Image interpretation reveals three possible flow paths from Apollinaris onto Gusev crater floor via a breach in Gusev s northwest rim; an unnamed crater that must have postdated any potential flow along the paths currently blocks the breach. After correcting for the unnamed crater, topographic profiles constructed along possible flow paths demonstrate that elevation along each path increases from 80 to 430 m. For two of the pathways, this elevation change indicates that lavas could not have traversed through them. The third profile, though, is less conclusive as the topography may reflect a broad flexural arch. However, on the basis of two of the topographic profiles together with the lack of obvious lava flow structures that indicate flow south through the breach, we propose that Apollinaris was not the direct source of the Gusev basalts. Instead, we argue that it seems more likely that the Gusev basalts were sourced from below the crater itself. The similar ages (as derived from crater counts) of Apollinaris s south flank and the Gusev plains lavas, however, suggest that the two volcanic regions may be part of a common magmatic feature. Citation: Lang, N. P., H. Y. McSween Jr., L. L. Tornabene, C. J. Hardgrove, and P. R. Christensen (2010), Reexamining the relationship between Apollinaris Patera and the basalts of the Gusev crater plains, Mars, J. Geophys. Res., 115,, doi: /2009je Introduction [2] Basaltic rocks analyzed on the plains of Gusev crater by the Spirit rover represent the most completely characterized suite of erupted materials on the Martian surface [McSween et al., 2004, 2006a]. These basalts are olivinerich (picritic) with mildly alkaline properties and appear to have formed from primitive magma produced by melting of an undepleted mantle at depth and then erupted without significant fractionation [Monders et al., 2007]. Their estimated viscosities range from 2.3 to 50 Pa s, comparable to lunar mare lavas and Archean high Mg basalts on Earth [Greeley et al., 2005]. Crater counting by Greeley et al. [2005] near the Spirit rover landing site yields a modeled age of 3.65 Ga for the Gusev plains. However, despite our 1 Department of Geology, Mercyhurst College, Erie, Pennsylvania, USA. 2 Department of Earth and Planetary Sciences and Planetary Geoscience Institute, University of Tennessee, Knoxville, Tennessee, USA. 3 Lunar and Planetary Laboratory, University of Arizona, Tucson, Arizona, USA. 4 Mars Space Flight Facility, School of Earth and Space Exploration, Arizona State University, Tempe, Arizona, USA. Copyright 2010 by the American Geophysical Union /10/2009JE range of knowledge regarding the Gusev basalts, their eruptive source remains ambiguous. Identifying the source of these basalts is important for further understanding their petrogenesis and for constraining the geologic and magmatic history of this region. [3] Some workers have proposed that the source of the Gusev basalts was the large ( 200 km diameter) volcano, Apollinaris Patera, located 200 km north of Gusev. Specifically, Martinez Alonso et al. [2005] suggest that lavas on the floor of Gusev may have flowed directly from Apollinaris Patera via a breach in Gusev s northwest corner. Such a scenario seems plausible given the low calculated viscosities of the Gusev basalts [Greeley et al., 2005] and the observation that some lavas on Mars have traveled for distances of >1000 km (e.g., in the Daedalia Planum region) [Keszthelyi et al., 2008; Lang et al., 2009]. Such a scenario is also bolstered by the Greeley et al. [2005] crater modeled ages of 3.76 Ga for Apollinaris s south flank deposits, which is similar to the 3.65 Ga modeled age for the Gusev plains. Despite the fact that no volcanic constructs or fissures inside Gusev have been identified to date, an alternative explanation is that the basalts were sourced from directly below the crater [Greeley et al., 2005; see also Milam et al., 2003; McSween et al., 2006a, 2006b]. In this scenario, the low calculated viscosities of the basalts might 1of20

2 explain the absence of obvious volcanic constructs within the crater [Greeley et al., 2005]. Here we test the hypothesis that the basaltic lavas on the plains of Gusev crater originated from Apollinaris Patera. Specifically, we integrate Thermal Emission Imaging System (THEMIS), Mars Orbiter Camera (MOC), High Resolution Imaging Science Experiment (HiRISE), and Mars Orbiter Laser Altimeter (MOLA) gridded data with a published geologic map of Kuzmin et al. [2000] to identify and assess plausible flow paths from Apollinaris Patera into Gusev crater. If basaltic lavas did flow from Apollinaris onto the floor of Gusev crater, then not only should the ages of the materials be similar, but there should have existed an unimpeded flow path connecting them. 2. Approach 2.1. Data Sets [4] Data sets used in this study were from the Mars Global Surveyor (MGS) ( ), Mars Odyssey (MO) (2001 to present), and Mars Reconnaissance Orbiter (MRO) (2005 to present) missions. MGS data included MOLA gridded digital elevation model (DEM) (1/128th degree gridded data set, which translates to 500 m/pixel) [Smith et al., 2001] and MOC imagery ( 3 m/pixel) ( moc_gallery/) [Malin and Edgett, 2001]; our motivation for using the gridded DEM rather than Precision Experiment Data Record (PEDR) tracks was that the gridded DEM provided more control in determining topography along possible lava flow paths (though some uncertainty is introduced; see sections 2.2 and 3.2.3) than individual PEDR tracks would. MO data included THEMIS visible (Vis, 18 m/pixel) as well as daytime infrared (IR) ( 100 m/pixel) imagery [Christensen et al., 2004]; THEMIS Vis images were obtained from the THEMIS instrument public Web site ( whereas THEMIS IR data were obtained from the Java based applet jmars ( edu/) [Gorelick et al., 2003; Weiss Malik et al., 2005]. MRO s HiRISE camera [McEwen et al., 2007] provided the highest resolution images ( 30 cm/pixel) that offered further morphologic context for MOC and THEMIS imagery Methodology [5] To test the hypothesis that the basalts on the plains of Gusev were directly sourced from Apollinaris Patera, we employed a multistep approach. First, we geologically and morphologically characterized the region from the south flank of Apollinaris Patera to the northern edge of Gusev crater, including the breach in Gusev s northwest corner, using THEMIS Vis and daytime IR data with MOC and HiRISE imagery combined with the published geologic map of Kuzmin et al. [2000]. This represents the region that would have been traversed by lava traveling from Apollinaris into Gusev. Hence, characterizing this region allowed us to identify potential Apollinaris Gusev flow paths. Normalized THEMIS daytime IR imagery obtained from jmars provided a seamless mosaic for use as a base map. We also attempted to use THEMIS IR data to help distinguish distinct spectral units in the study area. Unfortunately, Dust Cover Index (DCI) [Ruff and Christensen, 2002] values from 0.92 to 0.94 over this region indicate intermediate quantities of dust, which are sufficient to preclude spectral determinations. Therefore, our work builds mainly upon results and inferences from the Spirit rover mission in Gusev, as well as morphologic (i.e., THEMIS Vis, MOC, and HiRISE) and morphometric (i.e., MOLA) arguments. [6] We constructed topographic profiles along potential Apollinaris Gusev flow paths using the MOLA gridded DEM, taking into account an unnamed crater that impacted into the breach (Figure 1). In order to account for the excavation and deposition from this crater, we determined the transient (original) cavity diameter, depth of excavation, and the amount of rim uplift. We also determined the thickness of its ejecta blanket. We then subtracted the amount of uplift along the crater rim as well as the ejecta along each flow path profile in an attempt to reconstruct the preimpact topography along each path. This approach of accounting for the unnamed impact crater is outlined in more detail in section 3. Once the profiles were constructed and corrected for the effects of the unnamed impact crater, we were able to evaluate the potential for lava to have traveled from Apollinaris into Gusev. 3. Study 3.1. Geologic Framework and Flow Path Identification [7] Apollinaris Patera is a 200 km diameter volcano [Robinson et al., 1993; Scott et al., 1993] possibly constructed from both pyroclastic and effusive eruptions [Robinson et al., 1993] (Figure 1); crater modeled ages place Apollinaris Patera between 3.5 Ga [Robinson et al., 1993] and 3.75 Ga [Greeley et al., 2005], ages that place Apollinaris at around the Noachian Hesperian boundary [Hartmann, 2005]. Gusev is a 160 km diameter impact crater [Kuzmin et al., 2000] interpreted to have been the site of fluviolacustrine activity from the late Noachian to the mid Amazonian to late Amazonian [Cabrol et al., 1994; Cabrol and Brack, 1995; Cabrol and Grin, 1997; Kuzmin et al., 2000] but that appears to have also experienced volcanic activity [Milam et al., 2003; Martinez Alonso et al., 2005; McSween et al., 2004, 2006a, 2006b; Greeley et al., 2005]. From north to south, the terrain between Apollinaris and Gusev marks the transition from the northern lowlands to the southern highlands [Scott et al., 1993; Kuzmin et al., 2000] and is morphologically characterized by Apollinaris s south flank, an east west trending canyon system, and two distinct plateaus (P1 and P2 on Figure 1). [8] The south flank of Apollinaris Patera consists of a channeled, fan shaped deposit of layered materials (Figures 2a and 2b) that extend from a breach in the southern part of the construct s caldera[robinson et al., 1993; Scott et al., 1993]. On the basis of the proximity of these layered deposits to the volcano, we interpret the south flank materials to be volcanic; Greeley et al. [2005] model the flank deposits at 3.76 Ga. The distal edge of Apollinaris s south flank marks the northern boundary of Plateau 1 (P1; Figures 2a and 2c). Although much of the contact between these two features has been dissected and removed by the formation of the canyon system, it appears that at least some of the volcanic flank materials may have been emplaced on top of P1 (Figure 2c), suggesting that the south flank postdated P1 s formation. It also seems likely that because the canyon system cuts both the south flank and P1, canyon formation must also postdate P1 and volcanism at southern Apollinaris. 2of20

3 Figure 1. THEMIS daytime IR context map for the Apollinaris Patera Gusev crater region. Dashed white boxes labeled 3.76 Ga and 3.65 Ga represent the areas crater counted by Greeley et al. [2005]. Dashed black boxes show the locations of Figures 2a and 3a. P1 is Plateau 1; P2 is Plateau 2. Black star shows the landing location of the Spirit rover. THEMIS image obtained from jmars. [9] P1 forms an apparently planar surface at the 500 m contour interval (Figure 2a) that, on the basis of the apparent onlapping relationship of Apollinaris south flank materials onto P1, appears to be older than Apollinaris. Kuzmin et al. [2000] mapped P1 as a smooth surface composed of fluviolacustrine deposits associated with aqueous activity in Gusev crater. THEMIS Vis imagery illustrates P1 s smooth surface, which appears to be geologically homogeneous (at THEMIS Vis resolution), meaning that, with the exception of some impact craters and associated ejecta, P1 s surface 3of20

4 Figure 2 4of20

5 Figure 2. (continued) does not appear to be composed of multiple geologic units. Furthermore, except for a few north trending tectonic ridges (Figures 2a and 2d), P1 is devoid of tectonic structures. Also worth noting is that the available visible imagery of P1 does not reveal any surface textures consistent with primary volcanic morphologies, an observation consistent with the Kuzmin et al. [2000] interpretation for P1. [10] To its south, P1 has an abrupt transition to a second, topographically higher plateau (P2) that parallels the outside of Gusev s northern rim (Figures 2a, 2e, and 2f); P2 materials appear to be sitting on top of P1 materials, and we therefore interpret P2 materials to be younger than P1. Steps mark local transitions from P1 to P2. However, the overall transition is sharp, and at the resolution of available imagery, debris slopes separating the two morphologic features are absent (Figure 2f). Kuzmin et al. [2000] mapped P2 as volcanic deposits, an interpretation consistent with layers exposed in the walls of P2 (Figures 2g and 2h), although the possibility of P2 as Gusev related ejecta deposits cannot be excluded. The base of P2 is denoted mostly by the 1500 m contour, whereas the top locally reaches elevations above 1000 m (Figure 2a). Because P2 sits within this topographic interval, it is most likely older than Apollinaris s south flank materials. To elaborate, if P2 was emplaced after the materials on Apollinaris s south flank, it seems likely that the P2 materials would have extended across all of P1 and up Apollinaris s south flank to the 1500 m contour; this should especially be so if P2 materials accumulated and ponded along the outside of Gusev s northern rim (Figure 2a). However, this does not appear to have occurred. Although one could argue that P2 materials may have onlapped onto Apollinaris and were then eroded away (in fact, Figure 2. Imagery characterizing the terrain between Apollinaris Patera and Gusev crater. (a) Contoured (500 m spacing) THEMIS daytime IR context image of the terrain between Apollinaris Patera and Gusev crater. Dashed white boxes show the locations of Figures 2b 2h. P1 is Plateau 1, and P2 is Plateau 2. White arrows indicate the locations of some tectonic ridges that exist on P1. THEMIS image obtained from jmars; topography sourced from MOLA gridded DEM. (b) Portion of MOC image S of layers exposed in an impact crater on Apollinaris s south flank. On the basis of the proximity of these layers to Apollinaris, we interpret them to be volcanic in origin. (c) Portion of THEMIS Vis image V showing the contact between Apollinaris s south flank and P1 (highlighted by white arrows). Apollinaris s south flank appears to stand topographically higher than P1, suggesting that the flank materials were deposited on top of P1. (d) Portion of THEMIS Vis image V showing the smooth nature of the plateau s surface. With the exception of a few impact craters, this surface appears to be geologically homogeneous, meaning that it appears to consist of one geologic unit. Locally present are north trending tectonic ridges (highlighted by white arrows). (e) Portion of THEMIS Vis image V showing the characteristics of P2. P2 stands topographically higher than P1 and forms a bench that parallels the outside of Gusev s northern crater rim. (f) Portion of HiRISE image PSP_007737_1670 showing the nature of the boundary between P1 and P2 (highlighted by white arrows), which appears to be relatively sharp and locally marked by steps. (g) Portion of THEMIS Vis image V showing more characteristics of P2 and the impact crater that impacts P2. The walls of the impact crater reveal that P2 consists of apparently horizontal layers. The dashed white box shows the location of Figure 2h, and the dashed black box shows the location of Figure 3b. (h) Portion of MOC image R showing a higher resolution view of the P2 layers (highlighted by white arrows) exposed in walls of the unnamed crater. 5of20

6 Figure 2. P2 does appear to have experienced significant erosion), there is no evidence that Apollinaris s south flank has undergone extensive erosion similar to that of P2. In order for P2 to have been emplaced on Apollinaris and then eroded away, one would have to argue that extensive erosion on Apollinaris stopped once P2 was stripped away, a situation that does not seem geologically reasonable. Apollinaris s south flank appears too pristine (i.e., the flank channels seem to be too well preserved (Figure 2a)) to have been covered over and then subsequently reexposed by erosion. Therefore, we interpret that P2 must be older than Apollinaris s south flank. [11] Bounding P2 immediately to its south is Gusev s northern crater rim, which, together with P2, is breached (continued) near 12.9 S, E. This has resulted in an apparent pathway connecting P1 to the northern floor of Gusev crater; it is through this breach that Martinez Alonso et al. [2005] predicted that Apollinaris lavas flowed into Gusev crater. The floor of northern Gusev shows qualitatively fewer impact craters than P1; however, Kuzmin et al. [2000] mapped the floor as smooth material deposited by fluviolacustrine activity, similar to P1. Milam et al. [2003] noted that materials composing the floor of northern Gusev are layered and interpreted them to be consistent with fluviolacustrine, aeolian, or volcaniclastic in origin. Martinez Alonso et al. [2005] noted surface textures consistent with volcanic flows in various parts of northern Gusev. However, we have identified no morphologic features resembling volcanic 6of20

7 Figure 2. (continued) 7of20

8 Figure 2. (continued) 8of20

9 flows that suggest lava flows have entered into the crater through the breach. [12] If lava did flow into Gusev crater from the Apollinaris construct, then it must have flowed south, likely extending from Apollinaris s south flank, across P1 and through the breach of P2 and Gusev s northwest crater rim, a distance of 200 km. As mentioned in section 1, lavas traveling for such lengths on Mars are not unprecedented. Specifically, lava flows in the Daedalia Planum region have traveled south from Arsia Mons for distances of >1500 km [Keszthelyietal., 2008; Lang et al., 2009]; such lengths are not unique to Mars, as lava flows in Mare Imbrium on the Moon have traveled up to 1200 km, and in Sternia Fluctus on Venus, they have traveled up to 1000 km [Zimbelman, 1998]. Therefore, predicting that lava flows on Mars have traveled for 200 km is completely reasonable. Further, a closer examination of the breach (Figure 3a) reveals that it is characterized by three parallel, north trending valleys that are herein referred to as the western valley, middle valley, and eastern valley, respectively; thus, there are at least three paths that lava could have traveled. All three valleys appear to have been affected by a 22 km diameter unnamed crater located in P2; Figure 3a shows the crater and the extent of its ejecta as mapped by Kuzmin et al. [2000] with outlines of the three valleys. North of the unnamed impact crater, the western valley (Figure 3b) is an 18 km long, slightly sinuous trough that is 2 3 km wide; this valley (if preexisting) likely served as a conduit that funneled ejecta from the unnamed crater onto P1 [e.g., Kuzmin et al., 2000]. The western valley is truncated at 18 km by the unnamed crater, causing the valley to become obscured; we have extrapolated the valley s likely trend south of the crater and into Gusev. The middle valley (Figure 3c) is a 75 km long trough that varies from 1 to 7 km wide. The valley appears to follow along the rim of the unnamed crater and may have been disrupted by subsequent modification of the rim. Much of the floor of the middle valley hosts abundant blocks, and it is difficult to determine from Figure 3c whether they represent in place mesas or ejecta deposits. The eastern valley (Figure 3d) is 20 km wide and over 100 km long, making it the most prominent valley. On the basis of the mapping of Kuzmin et al. [2000], the eastern valley is partially blocked and infilled by ejecta from the unnamed crater; the eastern wall of the valley marks the boundary of the ejecta. Figure 3e is a HiRISE image of the southern end of the eastern valley and shows at least part of the ejecta consists of a thin (approximately meters to tens of meters thick) layer that may include blocks of what appears to be P2. These blocks may be similar to what is observed in Figure 3c in that they may represent in place outcrops of P2 and/or ejecta. On the basis of the trends of the western, middle, and eastern valleys, it seems plausible that the valleys merge into one at the northern edge of the Gusev crater floor. Because the unnamed crater truncates the western valley and ejecta partially fills the eastern valley, the unnamed crater likely postdates all three valleys Topographic Analysis [13] For lava to travel south along any of the flow paths, the route must have maintained a constant or decreasing elevation to the south. Therefore, a further test of the hypothesis that Apollinaris lavas flowed into Gusev is the construction of topographic profiles along the three possible flow paths. Any increase in elevation along any of the paths toward the south would have to be attributable to postvolcanic tectonism (including volcanic loading) or to the emplacement of a geologic unit subsequent to the flow of the lavas Topographic Profiles [14] Figure 4 presents the topographic profiles constructed using gridded MOLA data; the profiles extend from the east west trending canyon system, across P1, and through the breach in Gusev crater s northern rim. We constructed three topographic profiles labeled A A, B B, and C C, with each profile transecting one of the three identified valleys. For each profile, we pay particular attention to the topography and to the location of the cavity, rim, and ejecta of the unnamed crater. This is because the crater and ejecta interfere with all three pathways, making it unlikely that lava could have traveled any of the pathways subsequent to the emplacement of the unnamed crater. Therefore, lava traveling through the breach must predate the crater [Martinez Alonso et al., 2005]. In fact, on the basis of examination of available THEMIS, MOC, and HiRISE imagery, the ejecta represents the only recognizable geologic unit that postdates the lava. Subsequently, analysis of the potential for lava to travel along any of the three flow paths needs to account for the location of ejecta along each path with respect to elevation change. [15] The A A profile shows a slight increase in elevation ( 100 m) toward the south along P1 prior to intersection with the ejecta blanket (as mapped by Kuzmin et al. [2000]). Within the ejecta blanket, there is a 100 m topographic depression before the profile crosses P2. Along P2, topography increases to elevations above 1500 m, and examination of visible imagery here shows that some of this elevation increase is attributable to the uplifted rim of the unnamed crater. Elevation then decreases to below 2800 m where the profile crosses the unnamed crater. Once out of the cavity and beyond the rim, the profile exhibits a gradual decrease in elevation to the south toward the floor of Gusev crater. [16] The first 30 km of the B B profile show a slight decrease in elevation of 40 m along P1 before a sudden 80 m increase in elevation before intersecting with the ejecta blanket. As the profile, and hence flow path, nears the crater rim, the topography dramatically undulates; the cause of the undulation is most likely due to interaction of the profile with the abundant blocks present in the valley. However, as mentioned in section 3.1, it is unclear how many of these blocks are the result of subsequent landscape modification due to the unnamed crater. Moving away from the crater and onto the floor of Gusev crater, the topography becomes more gentle and the elevation levels out. [17] Similar to B B, the C C profile shows a slight decrease in elevation of 50 m along P1. However, at 25 km, there is a sudden 100 m increase in elevation that is followed by a broadly planar surface. At 70 km, there is an abrupt increase in elevation of 200 m followed by undulating topography that shows a gradual decrease to the south that becomes smoother on the northern floor of Gusev crater Topographic Profiles and the Unnamed Crater [18] Overall, the three topographic profiles in Figure 4 show elevation increases from Apollinaris toward Gusev. At first order, for lava to have traveled from Apollinaris to Gusev, such elevation changes require the lava to have flowed uphill. However, because the formation of the unnamed crater must postdate flow along the three pathways, could the 9of20

10 Figure 3a 10 of 20

11 Figure 3b. THEMIS Vis and MOLA data of the western valley. (right) Portion of THEMIS Vis image V showing the western valley. (top left) Gridded MOLA DEM showing the western valley location. (bottom left) Same as Figure 3b (top left), but without the location box. The white dashed hachured lines highlight the valley boundaries with the hachures pointing downslope. unnamed crater potentially account for all of the increases in elevation along each pathway? Specifically, could the ejecta be more extensive than mapped by Kuzmin et al. [2000]? Further, are all the high elevations surrounding the crater cavity (e.g., A A ) related to rim uplift associated with the cratering process? Answering these questions requires estimating the amount of uplift, excavation, and deposition that has occurred at the unnamed impact with respect to the topographic profiles so as to reconstruct and model the approximate topography prior to the emplacement of the unnamed impact. Estimating such factors is not straightforward and is difficult to quantify accurately [e.g., Stewart and Valiant, 2006]. However, because we are only attempting to approximate (to an order of magnitude) the extent to which the unnamed impact has contributed to the topography along each profile, we take a back of the envelope approach to constraining the unnamed crater s properties while noting possible caveats to our approach Volume Balance [19] We begin by quantifying the extent of the ejecta as mapped by Kuzmin et al. [2000] with the goal of making a rough volume estimate that can be compared to the volume Figure 3a. Imagery characterizing the breach in Gusev s northwest corner. (top) Contoured (500 m spacing) THEMIS daytime IR image of the breach and the unnamed crater that has subsequently blocked it. Dashed boxes show the locations of Figures 3b 3e. THEMIS daytime IR image obtained from jmars; topography sourced from MOLA gridded DEM. (bottom) Same region as the top image but with the impact ejecta as mapped by Kuzmin et al. [2000] shown. The labeled dashed black arrows highlight the three potential pathways that lava may have traveled from P1 onto the floor of Gusev. 11 of 20

12 Figure 3c. THEMIS Vis and MOLA data of the middle valley. (right) Portion of THEMIS Vis images V and V showing the middle valley. (top left) Gridded MOLA DEM showing the middle valley location. (bottom left) Same as Figure 3c (top left), but without the location box. The white dashed hachured lines highlight the valley boundaries with the hachures pointing downslope. of the unnamed crater s transient cavity. If the extent of the ejecta mapped by Kuzmin et al. [2000] is correct, then the volume of the ejecta should theoretically be on the same order of magnitude as the volume of the transient crater cavity. Calculating the volume of the crater cavity requires determining the original, or transient, diameter (D t ) of the crater as well as the depth of excavation (d e ). To calculate the transient diameter of the crater, we follow Grieve et al. [1981] and Melosh [1989] and use a conservative estimate: D t 0:5 to0:7d a ; where D a is the apparent, or current, diameter of the crater, which we measured from Kuzmin et al. [2000] to be 22 km (see Table 1). Because we are looking for an estimate of the ð1þ transient diameter, we use an average of the range (0.6), which results in a transient diameter of 13 km. [20] On the basis of Maxwell s Z model of excavation flow fields [Maxwell, 1977; Pike, 1980; Grieve et al., 1981; Melosh, 1989], the depth of excavation (d e ) is defined as d e 0:1D t ; which, using the value obtained from (1), results in a d e of 1.3 km, the approximate depth of the crater cavity as confirmed by MOLA and as seen in A A (Figure 4). However, the final depth of a transient cavity is a combination of both excavation and displacement of the target during the impact process, where only 1/3 of the crater cavity is attributed to ð2þ 12 of 20

13 Figure 3d. THEMIS Vis and MOLA data of the eastern valley. (right) Portion of THEMIS Vis images V and V showing the eastern valley. (top left) Gridded MOLA DEM showing the eastern valley location. (bottom left) Same as Figure 3d (top left), but without the location box. The white dashed hachured lines highlight the valley boundaries with the hachures pointing downslope. excavation and 2/3 are attributed to downward and outward displacement of the expanding transient cavity [e.g., Melosh, 1989]. Therefore, d e should not equal the final depth of the crater cavity even when accounting for postimpact modification including infilling and wall collapse. Subsequently, we take our calculated d e value as a gross overestimate of the excavation depth. [21] The volume of the transient unnamed crater (v c ) can be roughly estimated by approximating the crater as a spherical cap [Harris and Stocker, 1998]: v c ¼ ð=6þ 30:5D 2 t þ d 2 e de ; ð3þ which results in a crater volume of 90 km 3. Because the value of d e used in (3) is likely an overestimate, the value obtained from (3) is also likely an overestimate. [22] The ejecta volume (v b ) can be estimated following the methodology of McGetchin et al. [1973]: v b ¼ 2 t 0 ð0:5d t Þ 2 ; ð4þ where t 0, the thickness of the ejecta blanket at the transient crater rim, can be determined from [McGetchin et al., 1973] t 0 ¼ 0:14ð0:5D t Þ 0:74 ðr= ð0:5d t ÞÞ 3:0 ; ð5þ where r is the distance from the center of the transient crater to the point where the thickness is being calculated, which in this case is the crater rim. This means that in calculating the thickness of the ejecta at the crater rim, r represents the transient crater radius, which results in the second part of the 13 of 20

14 Figure 3e. Portion of HiRISE image PSP_007737_1670 showing a portion of the impact ejecta toward the southern end of the eastern valley (highlighted by white arrows). At least part of the ejecta blanket consists of a relatively thin (on the order of meters to tens of meters thick?) layer here. P2 locally outcrops here, but it is unclear if the blocks are in place and/or represent ejecta blocks. equation ((r/(0.5d t )) 3.0 ) reducing to 1. As such, (5) gives an ejecta thickness value of 93 m at the transient crater rim. [23] Using (4) and (5) results in an ejecta blanket volume of 25 km 3, which is approximately a factor of 4 less than the calculated crater volume. At first order, this seems to suggest that the ejecta may be more extensive than mapped by Kuzmin et al. [2000]. However, both volumes are the same order of magnitude, which, when paired with the possible gross overestimates that stemmed from our calculated d e value, makes us confident that the extent of the impact ejecta as mapped by Kuzmin et al. [2000] is reasonable Ejecta Thickness and Crater Cavity [24] Building upon the calculation of the crater cavity and ejecta volumes, each profile can be corrected for the crater cavity and ejecta blanket. To correct for the crater cavity requires (1) knowing the original, or transient (D t ), diameter of the impact crater, (2) knowing the thickness of the ejecta at the apparent diameter (D a ) of the crater rim, and (3) calculating the amount of uplift that has occurred along the crater rim; correcting for the crater cavity depth is critical for reconstructing the A A profile. Correcting for the ejecta requires calculating the thickness of the ejecta along each flow path; correcting for the ejecta is critical for reconstructing all three profiles. [25] We begin correcting for the crater cavity by estimating the amount of uplift and ejecta deposition that has occurred at the crater rim. On the basis of the unnamed crater s diameter and depth to diameter ratio of 0.045, we view this as a complex crater, leading us to use the empirical relationship for rim height for complex craters as a conservative estimate [Garvin et al., 2003]: h r ¼ 0:02D 0:84 a ; ð6þ which results in an h r value of 270 m at the original crater rim. This final rim height estimate is not entirely a consequence of uplift but is also influenced by [Melosh, 1989] (1) an overturned flap, (2) the augmentation of the rim from multiple breccia and impact melt dike injections into the transient crater wall, and (3) subsequent erosion of the crater rims with time. This uplift does not occur at just one single point at the rim crest but drops off to zero over a distance of 1.3 to 1.7 crater radii as measured from the transient crater rim [Melosh, 1989]. This amounts to a distance of 10 km for this situation of the unnamed crater. 14 of 20

15 Figure 4. Topographic profiles along each of the three flow paths. THEMIS daytime IR images showing the distribution of the ejecta [Kuzmin et al., 2000] from the unnamed crater and the locations of each of the three profiles. THEMIS daytime IR from jmars; topography data sourced from MOLA gridded data. [26] We use (5) to calculate the thickness of the ejecta blanket (t b ) at any point along the pathway as a function of distance from the crater center where t b substitutes for t 0 and r is the distance from the center of the transient cavity to the point where the thickness is being calculated. Table 2 lists the thickness of the ejecta blanket along each of the three pathways at 5 km increments as calculated using (3) and shows that the thicknesses vary from <1 to 20 m. [27] Assuming that the amount of uplift was the same around the crater rim, we can use our calculated value of h r plus the ejecta thickness at the rim of D a (Table 2) to estimate the preimpact surface elevation at the crater cavity for profile A A. Subtracting both h r and the ejecta thickness from the current rim of the crater should result in the preimpact elevations. However, to do so, we need to account for the effect of widening of the crater on the erasing of rim uplift; the widening of the crater is ultimately occurring because of blocks of the uplifted rim sliding down toward the crater floor, meaning that rim uplift is being erased by crater modification [Melosh, 1989]. To account for this widening and, thus, to determine the current lateral extent of rim uplift, we can take the difference between the transient crater radius ( 6.5 km) and the current crater radius ( 11 km) and then subtract that from 1.5D t ( 10 km); this amounts to uplift extending for 5.5 km from the current rim. Further, because the amount of uplift decreases as a function of distance from the transient crater rim, the amount of uplift present in the remaining 5.5 km should be less than the 270 m of uplift initially calculated as existing at the edge of the transient crater s rim (i.e., equation (6)). From Melosh [1989], it can be taken that this uplift may decrease as an exponential function. However, for simplicity in our calculations, we assume that uplift decreases linearly from the crater rim. Subsequently, if the amount of uplift that occurred at the transient crater s rim was 270 m and the uplift is 0 m at 10 km away, then, assuming a linear decrease, we can simply determine the amount of uplift in between. To illustrate, if uplift extends for 5.5 km from the current crater rim, then 4.5 km of uplifted rim have been removed, which would mean that there is a current uplift of 150 m at the crater s current rim. However, if uplift decreases exponentially from the crater rim as indicated by 15 of 20

16 Table 1. Values Used in Calculating Crater and Ejecta Parameters Symbol Meaning Value Source D a Apparent (final) crater diameter 22 km Measured from Kuzmin et al. [2000] D t Transient (initial) crater diameter 11 km Equation (1) d e Initial depth of crater excavation 1.3 km Equation (2) v c Impact crater volume 90 km 3 Equation (3) v b Ejecta blanket volume 25 km 3 Equation (4) p Pi 3.14 t 0 Ejecta thickness at the transient crater rim 93 m Equation (5) t b Ejecta thickness along a flow path See Table 2 Equation (5) r Distance from the center of the crater to a specific point in the ejecta blanket Measured from Kuzmin et al. [2000] h r Crater rim uplift 0.1 km Equation (6) Melosh [1989], then this value is likely an overcalculation. Further, the elevations of the crater rim represented in profile A A (Figure 4) are approximately 1400 m and approximately 1600 m. Subtracting 170 m (150 m of rim uplift plus 20 m of ejecta) from both crater rim elevations and extending it out one crater radii ( 5.5 km) results in elevations of 1570 m and 1770 m, respectively. The preimpact surface elevation should therefore reside between these two elevations such that the topography sloped south from approximately 1570 m to approximately 1770 m. Performing such a correction on the A A profile is critical because it passes directly through the impact crater; the pathway goes across the crater rim. In addition, although the B B profile does not pass through the crater, it does cross along the crater rim, meaning that it too needs to be corrected for rim uplift. However, the B B profile is at a distance of 5 km from the current crater rim, which translates to 14 m of uplift along B B. The C C profile is at a distance well beyond that affected by rim uplift, meaning that rim uplift does not need to be corrected for in that profile. [28] This approach to determining the preimpact surface elevation is similar to that described by Stewart and Valiant [2006]. Stewart and Valiant [2006] estimated the topography of the preimpact surface for Martian impacts by connecting a fraction of the crater rim with the surface beyond the ejecta blanket [see Stewart and Valiant 2006, Figure 2]; the preimpact surface should theoretically reside at the elevation that is beyond the ejecta blanket. However, because the surfaces along each of the three topographic profiles have some slopes to them (and are not flat as is assumed in the Stewart and Valiant [2006] approach), we cannot easily draw a line from the crater rim to beyond the ejecta blanket. Instead, we simply subtracted the estimated rim height and ejecta thickness and used the subsequent elevation as the preimpact surface elevation Corrected Profiles [29] Subsequently, each profile can be reconstructed, or corrected, on the basis of the values obtained from the methods described in sections and (Figure 5). [30] The corrected A A profile has the uplifted crater rim, the crater cavity, and the ejecta blanket removed. Removing the 170 m of rim uplift and ejecta on P2 results in surface elevations of approximately 1570 m and approximately 1770 m, which, as described in sections and , reveals the approximate preimpact surface. On the topographic profile, this results in P2 having a gradual slope to the south toward the floor of Gusev crater. Removal of the ejecta blanket shows minimal elevation change resulting in negligible change to the profile. Despite P2 s overall Table 2. Calculated Ejecta Thicknesses at 5 km Increments Along Each Topographic Profile Increment Distance From Crater Center (km) Calculated Ejecta Thickness (m) Western Valley (north rim of D a ) (south rim of D a ) Middle Valley Eastern Valley of 20

17 Figure 5. blanket. Topographic profiles shown in Figure 4, but corrected for the unnamed crater and its ejecta southerly slope, however, the profile still shows inclines of 100 and 430 m at 45 and 70 km, respectively. In addition, because we assumed a linear decrease in our accounting for rim uplift, we should note that this profile is likely an overcorrection of the real topography. This means that topography along A A may be higher than what is indicated in the profile. [31] The corrected B B profile shows corrections only for the impact ejecta. Similar to the corrected A A profile, there is negligible change to the profile with removal of the ejecta and rim uplift. Because it is unclear how much the blocks in the middle valley can be attributed to modification after the impact event, we have left the signal of that topography within the profile. Regardless, on P1 at 40 km, there is an 80 m increase in elevation. [32] The corrected C C profile has also only been corrected for the ejecta blanket, and similar to the B B profile, removal of the ejecta has had a negligible effect on the profile. There is still a notable two step increase in elevation at 25 and 70 km of 100 and 180 m, respectively Summary and Discussion [33] Even when correcting for the impact crater, uplifted rim, and ejecta, the topographic profiles along each possible flow path reveal that the surface across P1, and prior to reaching the breach of P2 and Gusev s northern rim, increases in elevation between 80 and 430 m. We should note, though, that because we are using gridded MOLA to construct our topographic profiles instead of individual PEDR tracks, there is most likely uncertainty within each of our constructed profiles. To elaborate, although the alongtrack shot spacing for MOLA is fairly constant at 300 m, the across track spacing is larger and irregular and can be as large as 4 km in the vicinity of the equator [Smith et al., 2001]. Because of this large and irregular grid spacing, not every grid cell will have a MOLA data point within it, meaning that there are interpolated values within the gridded data set. This suggests that we may be reaching the limits of resolution in the MOLA data presented here. In fact, some of the very short wavelength jaggedness observed in the topography presented in this paper (i.e., Figures 2a, 4, and 5) may reflect the resolution limit. That said, we believe that the overall trends in the topography presented here are accurate. For example, topographic undulations such as at 60 km in the A A profile (Figures 4 and 5), at 40 km in the B B profiles, and at 30 and 70 km in the C C profiles are all relatively gradual, occurring at length scales of 17 of 20

18 15 km. Such length scale changes in topography seem to suggest that we are resolving the overall topographic trends within each of the analyzed valleys. 4. Discussion [34] Comparison of the corrected topographic profiles with the geology described between Apollinaris and Gusev indicates that the increases in elevation cannot be attributed to obvious short wavelength faulting and/or folding [e.g., Kuzmin et al., 2000]. Although north trending tectonic ridges are present locally, no obvious tectonic structures are present that can be directly associated with topographic changes along each of the three flow paths (Figure 2a). It is possible that long wavelength topographic warping (lithospheric flexure) due to the loading of Apollinaris Patera could affect the topography along each profile [e.g., McGovern and Solomon, 1993]. In fact, the corrected A A profile (Figure 5) does seem to generally resemble a broad and gentle flexural arch profile, though a central divot does exist at 60 km. However, slight decreases in elevation toward the south in profiles B B and C C together with the multitiered elevation increases in the C C profile suggest that contributions to topography from lithospheric flexure are small compared to other sources of topography in both of those profiles; both B B and C C seem to show short wavelength sources of topography in the corrected profiles (Figure 5) (i.e., at 30 km in B B and 30 km and 70 km in C C ). Because the B B and C C profiles seem to show shorterwavelength topography sources, it is possible that the corrected A A profile (because of its close location to the B B and C C profiles) may have more complex shorter wavelength topography than what we have shown in Figure 5. In addition, on the basis of the analysis here using available imagery, the majority of the elevation increases cannot be attributed to any obvious geologic units emplaced subsequent to any lava flow into the northern part of Gusev. Together, these observations suggest that this rise in topography (at least in the B B and C C profiles) is mostly primary and indicate that if lavas did flow from Apollinaris into Gusev, they would have had to traverse m uphill. Specifically, for the case of the C C profile, lava would have had to traverse up two distinct rises. Climbing such inclines would require pressure driven flow for distances of 20 to 60 km up slopes of to Such a scenario seems unlikely for lava flows. In addition, for the A A profile, assuming that the divot at 60 km predates any hypothetical Apollinaris lava flow, the lava would had to have filled (and thereby erased) the central divot before continuing into Gusev crater. Therefore, on the basis of the assumption that the divot in the A A profile occurred before any hypothetical Apollinaris lava flowed across P1, we conclude that it is unlikely that Apollinaris sourced lavas could have flowed into Gusev crater. However, if the divot occurred after Apollinaris activity and we assume that the A A profile does reflect a flexural arch profile with no complex short wavelength topography (which may not be the case if the closely located B B and C C profiles show such effects) where south flowing Apollinaris lavas predated flexural arch formation, then there may have existed a more or less level pathway from Apollinaris into Gusev through the westernmost valley. If true, then we cannot robustly conclude that lavas from Apollinaris did not flow into Gusev crater [see Martinez Alonso et al., 2005]. That said, if south flowing Apollinaris lavas postdated, instead of predated, occurrence of the flexural arch, then lava flow into Gusev via the western valley would be unlikely as the flows would have to have traversed a 400 m hill. In addition, the lack of any obvious lava flow structures across the surface of P1 that indicate a flow direction to the south through the breach seems to suggest that lavas did not travel from the Apollinaris construct and into Gusev crater. Although P1 qualitatively exhibits fewer impact craters than the northern floor of Gusev crater, meaning that P1 could have been resurfaced subsequent to lava flow emplacement, the absence of obvious volcanic morphologies near the breach on Gusev s floor strongly suggests that lava did not flow into the crater via the breach. Therefore, we follow Greeley et al. [2005] and favor the alternative explanation that the Gusev basalts were locally erupted within the crater through fractures or from volcanic constructs that have been subsequently modified, destroyed, or buried. However, we do not discount the possibility of Apollinaris pyroclastic deposits existing within Gusev. Explosive eruptions most likely contributed to the construction of Apollinaris Patera [Robinson et al., 1993], and it seems likely that ash from those eruptions could have been deposited onto the floor of Gusev [e.g., Kerber et al., 2008; see also Milam et al., 2003]. [35] Even if the Gusev basalts are not sourced from the Apollinaris construct, the similar ages of Apollinaris s southern flank deposits and Gusev materials [Greeley et al., 2005] raise the possibility that the two are magmatically related [Greeley et al., 2005]. Perhaps Apollinaris magmatism and Gusev magmatism reflect different regions across a large plume. The mildly alkaline nature of the Gusev basalts [McSween et al., 2006b] might be consistent with a plume interpretation in which Gusev represents tapping of a plume margin where lower degrees of partial melting occur. In such a scenario, magma generated by the plume would rise through the crust with some magma possibly stalling and accumulating in magma chambers and eventually rising to the Gusev floor via preexisting fractures and zones of weakness that likely exist beneath (and were likely generated by) Gusev crater. This situation may be analogous to lunar mare lava emplacement where low viscosity basaltic lavas sourced from >100 km depth filled large impact basins over possible time scales of millions of years [Head, 1976; see also Greeley et al., 2005]. [36] A plume hypothesis may be testable through continued geologic mapping and spectroscopic studies across the region. Specifically, the identification of (likely buried?) regional fracture systems connecting Apollinaris and Gusev may be attributable to the interaction of a plume head on the base of the lithosphere [e.g., Ernst and Buchan, 2001]. Further, although the DCI [Ruff and Christensen, 2002] shows the Apollinaris region to be moderately dust covered and hence not suitable for spectroscopic analysis at THEMIS or Thermal Emission Spectrometer resolutions, the observation of layered deposits on the volcano s southern flank (Figure 2b) raises the possibility that future spectroscopic instruments may be able to analyze Apollinaris volcanic 18 of 20